Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App


In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the Jupyter Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this Jupyter notebook.


Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

Sample Dog Output

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

  • Step 0: Import Datasets
  • Step 1: Detect Humans
  • Step 2: Detect Dogs
  • Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
  • Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 5: Write your Algorithm
  • Step 6: Test Your Algorithm

Step 0: Import Datasets

Make sure that you've downloaded the required human and dog datasets:

  • Download the dog dataset. Unzip the folder and place it in this project's home directory, at the location /dogImages.

  • Download the human dataset. Unzip the folder and place it in the home directory, at location /lfw.

Note: If you are using a Windows machine, you are encouraged to use 7zip to extract the folder.

In the code cell below, we save the file paths for both the human (LFW) dataset and dog dataset in the numpy arrays human_files and dog_files.

In [1]:
import numpy as np
from glob import glob

# load filenames for human and dog images
human_files = np.array(glob("lfw/*/*"))
dog_files = np.array(glob("dogImages/*/*/*"))

# print number of images in each dataset
print('There are %d total human images.' % len(human_files))
print('There are %d total dog images.' % len(dog_files))
There are 13233 total human images.
There are 8351 total dog images.

Step 1: Detect Humans

In this section, we use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images.

OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory. In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

In [2]:
import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[0])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

In [3]:
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

  • What percentage of the first 100 images in human_files have a detected human face?
  • What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer: (You can print out your results and/or write your percentages in this cell)

In [4]:
from tqdm import tqdm

human_files_short = human_files[:100]
dog_files_short = dog_files[:100]

#-#-# Do NOT modify the code above this line. #-#-#

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.
num_humans = 0
num_dogs = 0

for f in human_files_short:
    if face_detector(f):
        num_humans+=1

for f in dog_files_short:
    if face_detector(f):
        num_dogs+=1

print(f'{num_humans}% human images files that were classified as humans(Sample of 100 images)')
print(f'There are {num_dogs}% dog images files that were classified as humans(Sample of 100 images)')
99% human images files that were classified as humans(Sample of 100 images)
There are 9% dog images files that were classified as humans(Sample of 100 images)

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [5]:
### (Optional) 
### TODO: Test performance of another face detection algorithm.
### Feel free to use as many code cells as needed.

Step 2: Detect Dogs

In this section, we use a pre-trained model to detect dogs in images.

Obtain Pre-trained VGG-16 Model

The code cell below downloads the VGG-16 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories.

In [6]:
import torch
import torchvision.models as models

# define VGG16 model
VGG16 = models.vgg16(pretrained=True)

# check if CUDA is available
use_cuda = torch.cuda.is_available()

# move model to GPU if CUDA is available
if use_cuda:
    VGG16 = VGG16.cuda()

Given an image, this pre-trained VGG-16 model returns a prediction (derived from the 1000 possible categories in ImageNet) for the object that is contained in the image.

(IMPLEMENTATION) Making Predictions with a Pre-trained Model

In the next code cell, you will write a function that accepts a path to an image (such as 'dogImages/train/001.Affenpinscher/Affenpinscher_00001.jpg') as input and returns the index corresponding to the ImageNet class that is predicted by the pre-trained VGG-16 model. The output should always be an integer between 0 and 999, inclusive.

Before writing the function, make sure that you take the time to learn how to appropriately pre-process tensors for pre-trained models in the PyTorch documentation.

In [7]:
from PIL import Image
import torchvision.transforms as transforms

# Set PIL to be tolerant of image files that are truncated.
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def VGG16_predict(img_path):
    '''
    Use pre-trained VGG-16 model to obtain index corresponding to 
    predicted ImageNet class for image at specified path
    
    Args:
        img_path: path to an image
        
    Returns:
        Index corresponding to VGG-16 model's prediction
    '''
    
    ## TODO: Complete the function.
    ## Load and pre-process an image from the given img_path
    ## Return the *index* of the predicted class for that image
    image = Image.open(img_path).convert('RGB')
    preprocess = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),])
    
    input_tensor = preprocess(image)
    input_batch = input_tensor.unsqueeze(0) # create a mini-batch as expected by the model

    # move the input and model to GPU for speed if available
    if torch.cuda.is_available():
        input_batch = input_batch.to('cuda')
        VGG16.to('cuda')

    with torch.no_grad():
        output = VGG16(input_batch)
    
    indices = (torch.nn.functional.softmax(output[0], dim=0)).argmax()
   
    return indices # predicted class index

(IMPLEMENTATION) Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained VGG-16 model, we need only check if the pre-trained model predicts an index between 151 and 268 (inclusive).

Use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

In [8]:
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    ## TODO: Complete the function.
    classification = VGG16_predict(img_path)
    if classification>=151 and classification<=268:
        return True
    else:
        return False

(IMPLEMENTATION) Assess the Dog Detector

Question 2: Use the code cell below to test the performance of your dog_detector function.

  • What percentage of the images in human_files_short have a detected dog?
  • What percentage of the images in dog_files_short have a detected dog?

Answer:

In [9]:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
num_humans = 0
num_dogs = 0

for f in human_files_short:
    if dog_detector(f):
        num_humans+=1

for f in dog_files_short:
    if dog_detector(f):
        num_dogs+=1

print(f'{num_humans}% human images files that were classified as Dogs(Sample of 100 images)')
print(f'There are {num_dogs}% dog images files that were classified as Dogs(Sample of 100 images)')
1% human images files that were classified as Dogs(Sample of 100 images)
There are 97% dog images files that were classified as Dogs(Sample of 100 images)

We suggest VGG-16 as a potential network to detect dog images in your algorithm, but you are free to explore other pre-trained networks (such as Inception-v3, ResNet-50, etc). Please use the code cell below to test other pre-trained PyTorch models. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [10]:
### (Optional) 
### TODO: Report the performance of another pre-trained network.
### Feel free to use as many code cells as needed.

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 10%. In Step 4 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have trouble distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador Chocolate Labrador Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively). You may find this documentation on custom datasets to be a useful resource. If you are interested in augmenting your training and/or validation data, check out the wide variety of transforms!

In [14]:
import os
from torchvision import datasets
import torchvision.transforms as transforms
from torch.utils.data.sampler import SubsetRandomSampler
import time
from torch.optim import lr_scheduler
from torchvision import datasets,models
%matplotlib inline
### TODO: Write data loaders for training, validation, and test sets
## Specify appropriate transforms, and batch_sizes
data_dir = "dogImages" 
batch_size = 20
num_workers = 0

transform = {
    'train': transforms.Compose([
        transforms.RandomResizedCrop(224),
        transforms.RandomHorizontalFlip(),
        transforms.RandomRotation(10),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ]),
    'valid': transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ]),
    'test': transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ]),    
}


# Preparing data loaders

image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x), transform[x])
                  for x in ['train', 'valid', 'test']}
    
loaders_scratch = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size = batch_size,
                                              shuffle = True, num_workers = num_workers)
                  for x in ['train', 'valid', 'test']}

class_names = image_datasets['train'].classes
n_classes = len(class_names)
In [15]:
def imshow(inp):
    inp = inp.numpy().transpose((1, 2, 0))
    mean = np.array([0.485, 0.456, 0.406])
    std = np.array([0.229, 0.224, 0.225])
    inp = std * inp + mean
    inp = np.clip(inp, 0, 1)    
    plt.imshow(inp)
    
# Get a batch of training data
images, classes = next(iter(loaders_scratch['train']))
      
fig = plt.figure(figsize=(25,4))
for idx in np.arange(20):
    ax = fig.add_subplot(2, 20/2, idx+1, xticks=[], yticks=[])
    imshow(images[idx])
    ax.set_title(class_names[classes[idx]].split(".")[1])

Question 3: Describe your chosen procedure for preprocessing the data.

  • How does your code resize the images (by cropping, stretching, etc)? What size did you pick for the input tensor, and why?
  • Did you decide to augment the dataset? If so, how (through translations, flips, rotations, etc)? If not, why not?

Answer:

  • I have resized my images to size 224 x 224 because that is what is the minimum requirement for pre-trained models. I have also used all 3 color channels(RGB) because this is another requirement.
  • For the training images I have performed flipping of images as well as random rotations(10 degrees). This helps us capture even cases where there are images of dogs in different angles. This also, prevents overfitting to a large extent. I have also implemented cropping of the images to extract just 224x224 out of the available 256x256 to reduce the effect of the background on the image.

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. Use the template in the code cell below.

In [16]:
%%markdown 
## I have implemented Data parallization to utilise all 8 of my GPUs on my AWS instance.(For the scratch as well as Transfer Learning)

I have implemented Data parallization to utilise all 8 of my GPUs on my AWS instance.(For the scratch as well as Transfer Learning)

In [17]:
import torch.nn as nn
import torch.nn.functional as F

# define the CNN architecture
class Net(nn.Module):
    ### TODO: choose an architecture, and complete the class
    def __init__(self):
        super(Net, self).__init__()
        
        ## Define layers of a CNN
        # convolutional layer (input 224x224x3 tensor)
        self.conv1 = nn.Conv2d(3, 16, 3, padding = 1 )        
        # convolutional layer (input 112x112x16 tensor)
        self.conv2 = nn.Conv2d(16, 32, 3, padding = 1)
        # convolutional layer (input 56x56x32 tensor)
        self.conv3 = nn.Conv2d(32, 64, 3, padding = 1)
        # convolutional layer (input 28x28x64 tensor)
        self.conv4 = nn.Conv2d(64, 128, 3, padding = 1)        
        # convolutional layer (input 14x14x128 tensor)
        self.conv5 = nn.Conv2d(128, 256, 3, padding = 1 )
        
        # max pooling layer
        self.pool = nn.MaxPool2d(2, 2)
        # dropout layer (p=0.2)
        self.dropout = nn.Dropout(0.2)
        #Batch Normalisation
        self.conv_bn1 = nn.BatchNorm2d(224,3)
        self.conv_bn2 = nn.BatchNorm2d(16)
        self.conv_bn3 = nn.BatchNorm2d(32)
        self.conv_bn4 = nn.BatchNorm2d(64)
        self.conv_bn5 = nn.BatchNorm2d(128)
        self.conv_bn6 = nn.BatchNorm2d(256)
        
        # linear layer (256 * 7 * 7 -> 512)
        self.fc1 = nn.Linear(256 * 7 * 7, 512)
        # linear layer (256 * 7 * 7 -> 133(Total number of classes))
        self.fc2 = nn.Linear(512, n_classes)
    
    def forward(self, x):
        ## Define forward behavior
        # add sequence of convolutional and max pooling layers
        x = self.conv_bn2(self.pool(F.relu(self.conv1(x))))
        x = self.conv_bn3(self.pool(F.relu(self.conv2(x)))) 
        x = self.conv_bn4(self.pool(F.relu(self.conv3(x))))
        x = self.conv_bn5(self.pool(F.relu(self.conv4(x))))
        x = self.conv_bn6(self.pool(F.relu(self.conv5(x))))
        
        # flatten image input
        x = x.view(-1, 256 * 7 * 7)        
        # add dropout layer
        x = self.dropout(x)
        # add second hidden layer
        x = F.relu(self.fc1(x))
        x = self.dropout(x)
        x = self.fc2(x)
        return x

#-#-# You so NOT have to modify the code below this line. #-#-#

# instantiate the CNN
model_scratch = Net()
print (model_scratch)

# check if CUDA is available
use_cuda = torch.cuda.is_available()
# move tensors to GPU if CUDA is available
if use_cuda:
    if torch.cuda.device_count() > 1:
        model_scratch = nn.DataParallel(model_scratch)
    model_scratch.cuda()
Net(
  (conv1): Conv2d(3, 16, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  (conv2): Conv2d(16, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  (conv3): Conv2d(32, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  (conv4): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  (conv5): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  (dropout): Dropout(p=0.2, inplace=False)
  (conv_bn1): BatchNorm2d(224, eps=3, momentum=0.1, affine=True, track_running_stats=True)
  (conv_bn2): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv_bn3): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv_bn4): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv_bn5): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv_bn6): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (fc1): Linear(in_features=12544, out_features=512, bias=True)
  (fc2): Linear(in_features=512, out_features=133, bias=True)
)

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step.

Answer: The steps in my CNN architecture are:

  • 5 convolutional layers. (Size = 3, stride = 1, padding = 1)
    1. conv1 => filter = 16
    2. conv2 => filter = 32
    3. conv3 => filter = 64
    4. conv4 => filter = 128
    5. conv5 => filter = 256
  • Each of the convulational layers are followed by a max pooling layer of 2*2 to down sample.
  • Each max pool is followed by a batch normalization allowing a higher learning rate.
  • The model is then reshaped to 25677 and then the dropout layer in used with a probability of dropout of 0.2 to avoid overfitting.
  • The last two layers of the model are linear layers, and the output is 133 nodes representing the 133 classes of dogs.

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_scratch, and the optimizer as optimizer_scratch below.

In [18]:
import torch.optim as optim

### TODO: select loss function
criterion_scratch = nn.CrossEntropyLoss()

### TODO: select optimizer
optimizer_scratch = optim.SGD(model_scratch.parameters(), lr=0.001, momentum=0.9)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_scratch.pt'.

In [19]:
# the following import is required for training to be robust to truncated images
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def train(n_epochs, loaders, model, optimizer, criterion, use_cuda, save_path):
    """returns trained model"""
    # initialize tracker for minimum validation loss
    valid_loss_min = np.Inf 
    
    for epoch in range(1, n_epochs+1):
        # initialize variables to monitor training and validation loss
        train_loss = 0.0
        valid_loss = 0.0
        
        ###################
        # train the model #
        ###################
        model.train()
        for batch_idx, (data, target) in enumerate(loaders['train']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## find the loss and update the model parameters accordingly
            ## record the average training loss, using something like
            ## train_loss = train_loss + ((1 / (batch_idx + 1)) * (loss.data - train_loss))
            
            # clear the gradients of all optimized variables
            optimizer.zero_grad()
            # forward pass: compute predicted outputs by passing inputs to the model
            output = model(data)
            # calculate the batch loss
            loss = criterion(output, target)
            # backward pass: compute gradient of the loss with respect to model parameters
            loss.backward()
            # perform a single optimization step (parameter update)
            optimizer.step()
            # update training loss
            train_loss = train_loss + (1 / (batch_idx + 1)) * (loss.data - train_loss)
            
        ######################    
        # validate the model #
        ######################
        model.eval()
        for batch_idx, (data, target) in enumerate(loaders['valid']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## update the average validation loss
            # forward pass: compute predicted outputs by passing inputs to the model
            output = model(data)
            # calculate the batch loss
            loss = criterion(output, target)
            # update average validation loss 
            valid_loss = valid_loss + (1 / (batch_idx + 1)) * (loss.data - valid_loss)

            
        # print training/validation statistics 
        print('Epoch: {} \tTraining Loss: {:.6f} \tValidation Loss: {:.6f}'.format(
            epoch, 
            train_loss,
            valid_loss
            ))
        
        ## TODO: save the model if validation loss has decreased
        if valid_loss <= valid_loss_min:
            print('Validation loss decreased ({:.6f} --> {:.6f}).  Saving model ...'.format(
            valid_loss_min,
            valid_loss))
            torch.save(model.state_dict(), save_path)
            valid_loss_min = valid_loss
            
    # return trained model
    return model
In [20]:
# train the model
n_epochs = 25
loaders_scratch = loaders_data
model_scratch = train(n_epochs, loaders_scratch, model_scratch, optimizer_scratch,
                      criterion_scratch, use_cuda, 'model_scratch.pt')
Epoch: 1 	Training Loss: 4.799298 	Validation Loss: 4.460819
Validation loss decreased (inf --> 4.460819).  Saving model ...
Epoch: 2 	Training Loss: 4.587278 	Validation Loss: 4.264039
Validation loss decreased (4.460819 --> 4.264039).  Saving model ...
Epoch: 3 	Training Loss: 4.470512 	Validation Loss: 4.134746
Validation loss decreased (4.264039 --> 4.134746).  Saving model ...
Epoch: 4 	Training Loss: 4.390939 	Validation Loss: 3.965838
Validation loss decreased (4.134746 --> 3.965838).  Saving model ...
Epoch: 5 	Training Loss: 4.295908 	Validation Loss: 3.954942
Validation loss decreased (3.965838 --> 3.954942).  Saving model ...
Epoch: 6 	Training Loss: 4.194137 	Validation Loss: 3.890123
Validation loss decreased (3.954942 --> 3.890123).  Saving model ...
Epoch: 7 	Training Loss: 4.139209 	Validation Loss: 3.964410
Epoch: 8 	Training Loss: 4.079586 	Validation Loss: 3.868158
Validation loss decreased (3.890123 --> 3.868158).  Saving model ...
Epoch: 9 	Training Loss: 4.018976 	Validation Loss: 3.789825
Validation loss decreased (3.868158 --> 3.789825).  Saving model ...
Epoch: 10 	Training Loss: 3.955300 	Validation Loss: 3.743257
Validation loss decreased (3.789825 --> 3.743257).  Saving model ...
Epoch: 11 	Training Loss: 3.901814 	Validation Loss: 3.642931
Validation loss decreased (3.743257 --> 3.642931).  Saving model ...
Epoch: 12 	Training Loss: 3.871402 	Validation Loss: 3.605811
Validation loss decreased (3.642931 --> 3.605811).  Saving model ...
Epoch: 13 	Training Loss: 3.782828 	Validation Loss: 3.629539
Epoch: 14 	Training Loss: 3.729008 	Validation Loss: 3.523917
Validation loss decreased (3.605811 --> 3.523917).  Saving model ...
Epoch: 15 	Training Loss: 3.700846 	Validation Loss: 3.560935
Epoch: 16 	Training Loss: 3.648127 	Validation Loss: 3.682963
Epoch: 17 	Training Loss: 3.593688 	Validation Loss: 3.527505
Epoch: 18 	Training Loss: 3.527639 	Validation Loss: 3.420984
Validation loss decreased (3.523917 --> 3.420984).  Saving model ...
Epoch: 19 	Training Loss: 3.518352 	Validation Loss: 3.533019
Epoch: 20 	Training Loss: 3.453376 	Validation Loss: 3.581699
Epoch: 21 	Training Loss: 3.394987 	Validation Loss: 3.388971
Validation loss decreased (3.420984 --> 3.388971).  Saving model ...
Epoch: 22 	Training Loss: 3.371547 	Validation Loss: 3.303031
Validation loss decreased (3.388971 --> 3.303031).  Saving model ...
Epoch: 23 	Training Loss: 3.364148 	Validation Loss: 3.279624
Validation loss decreased (3.303031 --> 3.279624).  Saving model ...
Epoch: 24 	Training Loss: 3.321362 	Validation Loss: 3.308079
Epoch: 25 	Training Loss: 3.266993 	Validation Loss: 3.257376
Validation loss decreased (3.279624 --> 3.257376).  Saving model ...

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 10%.

In [21]:
def test(loaders, model, criterion, use_cuda):

    # monitor test loss and accuracy
    test_loss = 0.
    correct = 0.
    total = 0.

    model.eval()
    for batch_idx, (data, target) in enumerate(loaders['test']):
        # move to GPU
        if use_cuda:
            data, target = data.cuda(), target.cuda()
        # forward pass: compute predicted outputs by passing inputs to the model
        output = model(data)
        # calculate the loss
        loss = criterion(output, target)
        # update average test loss 
        test_loss = test_loss + ((1 / (batch_idx + 1)) * (loss.data - test_loss))
        # convert output probabilities to predicted class
        pred = output.data.max(1, keepdim=True)[1]
        # compare predictions to true label
        correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())
        total += data.size(0)
            
    print('Test Loss: {:.6f}\n'.format(test_loss))

    print('\nTest Accuracy: %2d%% (%2d/%2d)' % (
        100. * correct / total, correct, total))
In [22]:
# call test function    
test(loaders_scratch, model_scratch, criterion_scratch, use_cuda)
Test Loss: 3.320138


Test Accuracy: 23% (194/836)

Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively).

If you like, you are welcome to use the same data loaders from the previous step, when you created a CNN from scratch.

In [23]:
## TODO: Specify data loaders
# Using old data loaders.

(IMPLEMENTATION) Model Architecture

Use transfer learning to create a CNN to classify dog breed. Use the code cell below, and save your initialized model as the variable model_transfer.

In [24]:
import torchvision.models as models
import torch.nn as nn

## TODO: Specify model architecture 
model_transfer = models.densenet161(pretrained=True)
for param in model_transfer.parameters():
    param.requires_grad = False
num_ftrs = model_transfer.classifier.in_features
model_transfer.classifier = nn.Linear(num_ftrs, n_classes)

# if GPU is available, move the model to GPU
if use_cuda:
    if torch.cuda.device_count() > 1:
        model_transfer = nn.DataParallel(model_transfer)
    model_transfer.cuda()
print(model_transfer)
DataParallel(
  (module): DenseNet(
    (features): Sequential(
      (conv0): Conv2d(3, 96, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
      (norm0): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu0): ReLU(inplace=True)
      (pool0): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (denseblock1): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(96, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(144, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(192, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(240, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(288, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(336, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(336, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition1): _Transition(
        (norm): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock2): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(192, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(240, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(288, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(336, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(336, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(432, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(432, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(480, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(528, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(528, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(576, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(624, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(624, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(720, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(720, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition2): _Transition(
        (norm): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(768, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock3): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(432, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(432, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(480, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(528, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(528, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(576, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(624, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(624, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(720, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(720, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(816, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(816, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(864, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(912, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(912, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer13): _DenseLayer(
          (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(960, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer14): _DenseLayer(
          (norm1): BatchNorm2d(1008, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1008, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer15): _DenseLayer(
          (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1056, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer16): _DenseLayer(
          (norm1): BatchNorm2d(1104, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1104, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer17): _DenseLayer(
          (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer18): _DenseLayer(
          (norm1): BatchNorm2d(1200, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1200, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer19): _DenseLayer(
          (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1248, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer20): _DenseLayer(
          (norm1): BatchNorm2d(1296, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1296, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer21): _DenseLayer(
          (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1344, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer22): _DenseLayer(
          (norm1): BatchNorm2d(1392, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1392, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer23): _DenseLayer(
          (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1440, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer24): _DenseLayer(
          (norm1): BatchNorm2d(1488, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1488, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer25): _DenseLayer(
          (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1536, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer26): _DenseLayer(
          (norm1): BatchNorm2d(1584, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1584, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer27): _DenseLayer(
          (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1632, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer28): _DenseLayer(
          (norm1): BatchNorm2d(1680, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1680, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer29): _DenseLayer(
          (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1728, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer30): _DenseLayer(
          (norm1): BatchNorm2d(1776, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1776, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer31): _DenseLayer(
          (norm1): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1824, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer32): _DenseLayer(
          (norm1): BatchNorm2d(1872, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1872, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer33): _DenseLayer(
          (norm1): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1920, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer34): _DenseLayer(
          (norm1): BatchNorm2d(1968, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1968, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer35): _DenseLayer(
          (norm1): BatchNorm2d(2016, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2016, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer36): _DenseLayer(
          (norm1): BatchNorm2d(2064, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2064, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (transition3): _Transition(
        (norm): BatchNorm2d(2112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (relu): ReLU(inplace=True)
        (conv): Conv2d(2112, 1056, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
      )
      (denseblock4): _DenseBlock(
        (denselayer1): _DenseLayer(
          (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1056, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer2): _DenseLayer(
          (norm1): BatchNorm2d(1104, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1104, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer3): _DenseLayer(
          (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer4): _DenseLayer(
          (norm1): BatchNorm2d(1200, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1200, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer5): _DenseLayer(
          (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1248, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer6): _DenseLayer(
          (norm1): BatchNorm2d(1296, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1296, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer7): _DenseLayer(
          (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1344, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer8): _DenseLayer(
          (norm1): BatchNorm2d(1392, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1392, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer9): _DenseLayer(
          (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1440, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer10): _DenseLayer(
          (norm1): BatchNorm2d(1488, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1488, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer11): _DenseLayer(
          (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1536, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer12): _DenseLayer(
          (norm1): BatchNorm2d(1584, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1584, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer13): _DenseLayer(
          (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1632, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer14): _DenseLayer(
          (norm1): BatchNorm2d(1680, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1680, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer15): _DenseLayer(
          (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1728, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer16): _DenseLayer(
          (norm1): BatchNorm2d(1776, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1776, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer17): _DenseLayer(
          (norm1): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1824, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer18): _DenseLayer(
          (norm1): BatchNorm2d(1872, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1872, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer19): _DenseLayer(
          (norm1): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1920, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer20): _DenseLayer(
          (norm1): BatchNorm2d(1968, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(1968, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer21): _DenseLayer(
          (norm1): BatchNorm2d(2016, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2016, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer22): _DenseLayer(
          (norm1): BatchNorm2d(2064, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2064, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer23): _DenseLayer(
          (norm1): BatchNorm2d(2112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2112, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
        (denselayer24): _DenseLayer(
          (norm1): BatchNorm2d(2160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu1): ReLU(inplace=True)
          (conv1): Conv2d(2160, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu2): ReLU(inplace=True)
          (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        )
      )
      (norm5): BatchNorm2d(2208, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (classifier): Linear(in_features=2208, out_features=133, bias=True)
  )
)

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer:

For this problem I used transfer learning as the solution and the base model implemented was the dense net. To implement this model I used all the layers of the dense net model except the last one because the number of classes that we need to classify are specific to our problem statement. (133 classes)

This is suitable to this problem because of the lack of training data made available to me. I only have 6600 images to classify 133 classes. Therefore using the frozen weights from the trained layers of the densenet model, we already get quite some pattern recognition ability inherant to the densenet model. We then using our training data to better the model for our specific use case which is dog breed classification.

I used the learnings from this repository [https://github.com/bamos/densenet.pytorch] for my understanding and the implementation of the Densenet model.

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_transfer, and the optimizer as optimizer_transfer below.

In [25]:
criterion_transfer = nn.CrossEntropyLoss()
#Using model_transfer.module because I am using multiple GPU
#Source:https://discuss.pytorch.org/t/how-to-reach-model-attributes-wrapped-by-nn-dataparallel/1373
optimizer_transfer = optim.SGD(model_transfer.module.classifier.parameters(), lr=0.001, momentum=0.9)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_transfer.pt'.

In [26]:
# train the model
n_epochs = 25
loaders_transfer = loaders_scratch
model_transfer = train(n_epochs, loaders_transfer, model_transfer, optimizer_transfer, criterion_transfer, use_cuda, 'model_transfer.pt')
Epoch: 1 	Training Loss: 4.651134 	Validation Loss: 3.850690
Validation loss decreased (inf --> 3.850690).  Saving model ...
Epoch: 2 	Training Loss: 4.121881 	Validation Loss: 3.025498
Validation loss decreased (3.850690 --> 3.025498).  Saving model ...
Epoch: 3 	Training Loss: 3.709174 	Validation Loss: 2.401643
Validation loss decreased (3.025498 --> 2.401643).  Saving model ...
Epoch: 4 	Training Loss: 3.360073 	Validation Loss: 2.017637
Validation loss decreased (2.401643 --> 2.017637).  Saving model ...
Epoch: 5 	Training Loss: 3.101313 	Validation Loss: 1.573489
Validation loss decreased (2.017637 --> 1.573489).  Saving model ...
Epoch: 6 	Training Loss: 2.885979 	Validation Loss: 1.340911
Validation loss decreased (1.573489 --> 1.340911).  Saving model ...
Epoch: 7 	Training Loss: 2.734776 	Validation Loss: 1.240106
Validation loss decreased (1.340911 --> 1.240106).  Saving model ...
Epoch: 8 	Training Loss: 2.585925 	Validation Loss: 1.160881
Validation loss decreased (1.240106 --> 1.160881).  Saving model ...
Epoch: 9 	Training Loss: 2.438297 	Validation Loss: 0.950522
Validation loss decreased (1.160881 --> 0.950522).  Saving model ...
Epoch: 10 	Training Loss: 2.347317 	Validation Loss: 0.892168
Validation loss decreased (0.950522 --> 0.892168).  Saving model ...
Epoch: 11 	Training Loss: 2.262348 	Validation Loss: 0.856458
Validation loss decreased (0.892168 --> 0.856458).  Saving model ...
Epoch: 12 	Training Loss: 2.174391 	Validation Loss: 0.823745
Validation loss decreased (0.856458 --> 0.823745).  Saving model ...
Epoch: 13 	Training Loss: 2.120696 	Validation Loss: 0.731483
Validation loss decreased (0.823745 --> 0.731483).  Saving model ...
Epoch: 14 	Training Loss: 2.082058 	Validation Loss: 0.664457
Validation loss decreased (0.731483 --> 0.664457).  Saving model ...
Epoch: 15 	Training Loss: 2.019109 	Validation Loss: 0.665576
Epoch: 16 	Training Loss: 1.981707 	Validation Loss: 0.636074
Validation loss decreased (0.664457 --> 0.636074).  Saving model ...
Epoch: 17 	Training Loss: 1.916068 	Validation Loss: 0.639430
Epoch: 18 	Training Loss: 1.889228 	Validation Loss: 0.575583
Validation loss decreased (0.636074 --> 0.575583).  Saving model ...
Epoch: 19 	Training Loss: 1.865845 	Validation Loss: 0.588830
Epoch: 20 	Training Loss: 1.819650 	Validation Loss: 0.567095
Validation loss decreased (0.575583 --> 0.567095).  Saving model ...
Epoch: 21 	Training Loss: 1.785024 	Validation Loss: 0.552936
Validation loss decreased (0.567095 --> 0.552936).  Saving model ...
Epoch: 22 	Training Loss: 1.780925 	Validation Loss: 0.513912
Validation loss decreased (0.552936 --> 0.513912).  Saving model ...
Epoch: 23 	Training Loss: 1.755021 	Validation Loss: 0.508403
Validation loss decreased (0.513912 --> 0.508403).  Saving model ...
Epoch: 24 	Training Loss: 1.755715 	Validation Loss: 0.534587
Epoch: 25 	Training Loss: 1.699042 	Validation Loss: 0.471028
Validation loss decreased (0.508403 --> 0.471028).  Saving model ...

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 60%.

In [27]:
# load the model that got the best validation accuracy (uncomment the line below)
#model_transfer.load_state_dict(torch.load('model_transfer.pt'))
In [28]:
test(loaders_transfer, model_transfer, criterion_transfer, use_cuda)
Test Loss: 0.507572


Test Accuracy: 84% (704/836)

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan hound, etc) that is predicted by your model.

In [29]:
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
import matplotlib.image as mpimg
from torch.autograd import Variable

class_names = [item[4:].replace("_", " ") for item in image_datasets['train'].classes]

def predict_breed_transfer(img_path):
    
    # load the image and return the predicted breed    
    img = Image.open(img_path).convert('RGB') # Load the image from provided path
    
    
    normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                     std=[0.229, 0.224, 0.225])
    preprocess = transforms.Compose([transforms.Resize(224),
                                     transforms.CenterCrop(224),
                                     transforms.ToTensor(),
                                     normalize])
    img_tensor = preprocess(img).float()
    img_tensor.unsqueeze_(0)  # Insert the new axis at index 0 i.e. in front of the other axes/dims.
    img_tensor = Variable(img_tensor) #The input to the network needs to be an autograd Variable
    if use_cuda:
        img_tensor = Variable(img_tensor.cuda())        
    model_transfer.eval()
    output = model_transfer(img_tensor) # Returns a Tensor of shape (batch, num class labels)
    output = output.cpu()
    predict_index = output.data.numpy().argmax() # Our prediction will be the index of the class label with the largest value.
    return predict_index, class_names[predict_index], image_datasets['train'].classes[predict_index]
In [30]:
### Display prediction
def display_predictions(img_path):
    #print (img_path)
    pred_index, breed, name = predict_breed_transfer(img_path)
    print("This is a dog!")
        
    # display test image
    fig = plt.figure(figsize=(16,4))
    ax = fig.add_subplot(1,2,1)
    img = mpimg.imread(img_path)
    ax.imshow(img)
    plt.axis('off')

    # display sample of matching breed images
    subdir = '/'.join(['dogImages/valid', str(name)])
    file = random.choice(os.listdir(subdir))
    path = '/'.join([subdir, file])
    ax = fig.add_subplot(1,2,2)
    img = mpimg.imread(path)
    ax.imshow(img.squeeze(), cmap="gray", interpolation='nearest')
    plt.title(breed)
    plt.axis('off')
    plt.show()   
    
    # extract breed from image path
    actual_breed = img_path.split('/')[3].split('.')[0]
    print(f"Actual Breed: {actual_breed}\n")
    print(f"Predicted Breed: {breed}\n")
    print("\n"*3)
In [31]:
# Create list of test image paths
import random
test_img_paths = sorted(glob('dogImages/test/*/*'))
# Shuffle the list and display first few rows
np.random.shuffle(test_img_paths)
test_img_paths[1:5]

for img_path in test_img_paths[0:5]:
    display_predictions(img_path)
This is a dog!
Actual Breed: Cane_corso_03114

Predicted Breed: Boxer





This is a dog!
Actual Breed: Old_english_sheepdog_07390

Predicted Breed: Bearded collie





This is a dog!
Actual Breed: Havanese_05631

Predicted Breed: Havanese





This is a dog!
Actual Breed: Papillon_07440

Predicted Breed: Papillon





This is a dog!
Actual Breed: Australian_terrier_00923

Predicted Breed: Silky terrier






Step 5: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

  • if a dog is detected in the image, return the predicted breed.
  • if a human is detected in the image, return the resembling dog breed.
  • if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 4 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

Sample Human Output

(IMPLEMENTATION) Write your Algorithm

In [32]:
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

def run_app(img_path):
    '''
    Use pre-trained model to to check if the image at the given path
    contains a human being or a dog or none. 
    
    Args:
        img_path: path to an image
        
    Returns:
        print if a human face is detected or not
        print the dog breed or show that neither human face nor a dog detected 
    '''            
    is_human = face_detector(img_path)
    is_dog = dog_detector(img_path)
    pred_index, breed, name = predict_breed_transfer(img_path)
        
    # display test image
    fig = plt.figure(figsize=(16,4))
    
    if(is_human):
        print("This is a human!")
        ax = fig.add_subplot(1,2,1)
        img = mpimg.imread(img_path)
        ax.imshow(img)
        plt.axis('off')

        # display sample of matching breed images
        subdir = '/'.join(['dogImages/valid', str(name)])
        file = random.choice(os.listdir(subdir))
        path = '/'.join([subdir, file])
        ax = fig.add_subplot(1,2,2)
        img = mpimg.imread(path)
        ax.imshow(img.squeeze(), cmap="gray", interpolation='nearest')
        plt.title(breed)
        plt.axis('off')
        plt.show()   
        print("This looks like a -> " + breed)
        print("\n"*3)
        return
    
    elif(is_dog):
        print("This is a dog!")
        ax = fig.add_subplot(1,2,1)
        img = mpimg.imread(img_path)
        ax.imshow(img)
        plt.axis('off')

        # display sample of matching breed images
        subdir = '/'.join(['dogImages/valid', str(name)])
        file = random.choice(os.listdir(subdir))
        path = '/'.join([subdir, file])
        ax = fig.add_subplot(1,2,2)
        img = mpimg.imread(path)
        ax.imshow(img.squeeze(), cmap="gray", interpolation='nearest')
        plt.title(breed)
        plt.axis('off')
        plt.show()   
        print("You look like ... " + breed)
        print("\n"*3)
        return
    
    else:
        print('Error!... I can not determine what you are!')
        ax = fig.add_subplot(1,2,1)
        img = mpimg.imread(img_path)
        ax.imshow(img)
        plt.axis('off')
        plt.show()    
        print("\n"*3)
        return

Step 6: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: (Three possible points for improvement) The achieved accuracy was 83% which is pretty good. The model works well if the image has either a dog or a human. The possible points of improvements are:

  • A larger training dataset could be used to get better results
  • A better learning rate might be useful.
  • Investigating other pre-trained models with a better performance could give better results.
In [33]:
## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.

## suggested code, below
for file in np.hstack((human_files[:3], dog_files[:3])):
    run_app(file)
This is a human!
This looks like a -> Boston terrier




This is a human!
This looks like a -> Irish wolfhound




This is a human!
This looks like a -> Field spaniel




This is a dog!
You look like ... Norfolk terrier




This is a dog!
You look like ... Norfolk terrier




This is a dog!
You look like ... Norfolk terrier




In [34]:
test_files = np.array(glob("images/*.jpg"))

# print number of images in each dataset
print('There are %d total test images' % len(test_files))

for file in np.hstack((test_files[:])):
    print(file)
    run_app(file)
There are 7 total test images
images/Welsh_springer_spaniel_08203.jpg
This is a dog!
You look like ... Welsh springer spaniel




images/Labrador_retriever_06457.jpg
This is a dog!
You look like ... Labrador retriever




images/Brittany_02625.jpg
This is a dog!
You look like ... Brittany




images/Labrador_retriever_06455.jpg
This is a dog!
You look like ... Chesapeake bay retriever




images/Labrador_retriever_06449.jpg
This is a dog!
You look like ... Labrador retriever




images/American_water_spaniel_00648.jpg
This is a dog!
You look like ... Boykin spaniel




images/Curly-coated_retriever_03896.jpg
This is a dog!
You look like ... Curly-coated retriever




In [ ]: